Getting Started: Adopt the Pipeline¶

Prerequisites

Cursor installed with agent models enabled
Python 3.10+, Node.js 20+, git
Models enabled: gemini-3.1-pro, gpt-5.3-codex, gpt-5.1-codex-max, gemini-3-flash, claude-4.6-sonnet (check Cursor Settings > Models)

Choose your tier¶

Most teams: Start with Practitioner (8 agents, full pipeline)
Advanced teams with cost concerns: Use Expert (tiered routing, ~15 agents)
New to AI agents: Read Foundation first (concepts only)

Copy into your project¶

cp -r .cursor-practitioner/* your-project/.cursor/

Customize for your team¶

Use team- prefix for team-specific agents/rules
Keep jg- files as read-only upstream references
Diff before overwriting if you've modified any jg- files

Run your first real issue¶

Create or pick an issue in your project
Paste in Cursor: "Work on issue #N: [description]"
The jg-planner-first rule triggers automatic delegation
Watch the pipeline: planner → subplanner → worker → tester → reviewer → git
Review the generated PR

What "done" looks like¶

The pipeline produces a PR with conventional commits, passing tests, and a review-result. A human reviews and merges.

Upgrading tiers¶

Copy the new tier directory over .cursor/
Keep team- prefixed files (they survive upgrades)
Diff before overwriting modified jg- files

Troubleshooting¶

Agent didn't pick up rule

Check frontmatter, description field, and file location. The file must be in .cursor/rules/, have valid frontmatter (between --- markers), and the description field must accurately describe when it should apply.

Model not found

Enable the model in Cursor Settings > Models. Some models are hidden by default and must be enabled before agents can use them.

Pipeline artifacts not appearing

The first agent creates .pipeline/<issue-id>/. If you're starting fresh, the planner creates this directory.